Given the development of multimedia communication and the increasingly powerful functions of computers, image processing of dynamic objects has gradually become a popular topic for research. The images of dynamic objects are usually displayed as full-motion videos by displaying serial images. Accordingly, serial image processing has become an important technique for images processing.
Since serial images involve a series of images, identifying the target objects shown in the images is a main task in images processing. In practice, the targets of serial images are divided, identified and traced. The technique applies to many fields, such as automatic monitoring, sports broadcasting, video frequency compression, monitoring production lines and intelligent traffic control.
In addition, target tracing in images processing has become a key technique in the latest technology of MPEG-4, wherein MPEG-4z provides content-based image operation by means of the technique of target tracing and division. Therefore, a thorough exploration of this technology will have far-reaching effects upon the development of its various applications.
For the time being, there are many computation methods for target identification and tracing, wherein the most widely used is the computation method of pattern matching and updating that is based on the features of images. This computation method compares the image features currently extracted with the features of the pattern of the target being traced. The features adopted are usually the marginal features of the images. The last step of the computation method is matching the target pattern and the current images, and updating the pattern simultaneously. A recently developed means of pattern matching based on Hausdorff distance is quite effective. However, the disadvantage of this is that, in the course of matching, the space of searching increases rapidly with the size of the image and the size of the pattern, creating a larger and larger workload for the memory and the processor required to conduct the searching. Furthermore, extracting image features is not an easy job, as the efficiency of searching and tracing is reduced with the size of image and the size of pattern, making goals achievement extremely difficult.
Therefore, the most important issues of research are precision, automation and efficiency in target tracing.